Skip to main content

14 Robots Teach Themselves to Walk Into a Bar

Can robots be trained to pick up objects they've never seen before?  Recent research indicates that the best approach may be to allow robots to experiment on their own, and then share that knowledge with other robots.  

How To Train Your Robot

Until now, robots were programmed to recognize specific objects and to relate to those objects in a pre-programmed fashion. This is a good approach for predictable tasks, in planned locations, with previously experienced objects. In the real world, life is what happens while you’re making other plans.
At Google, Sergey Levine and his team took fourteen robotic arms, networked them together, and used (CNN) convolutional neural networks to let these robots learn on their own how to pick up small objects including a cup, tape dispenser, and a toy dolphin.
Over time, and nearly one million practice runs, the robots were able to self-correct and optimize their actions.  They learned to pick up objects faster and more often, employing  self-taught strategies - such as pushing an object away in order to reach another, and developed different techniques to pick up soft versus hard objects.

The robots learned from practice and sharing knowledge with one another, and without human interaction or pre-programming. In fact, the only human contribution was to replace the objects in the trays. 

During the study, the robots were positioned in front of trays with randomly placed objects varying in shape, scale, weight and hardness. Each of the 14 robot arms had a similar, but not identical design. 
Each robot’s successes and failures were fed to a shared convoluted neural network, a “robo-wiki” of shared knowledge accessible to the 14 robots. 
“Our experimental evaluation demonstrates that our method achieves effective real-time control, can successfully grasp novel objects, and corrects mistakes by continuous robotic hand-eye coordination...”  
Google’s project builds on the concept that robots can teach other robots how to perform specific tasks—or, more accurately, how one artificial intelligence can help another AI improve, without a human ever intervening. 
Here's a link to the Google Research: Learning Hand-Eye Coordination for Robotic Grasping with Deep Learning and Large-Scale Data Collection - Sergey Levine, Peter Pastor, Alex Krizhevsky, and Deirdre Quillen

The robots taught each other to pick up hard and soft objects in different ways, without human guidance. 

For an object that was perceived as rigid the robots would just grasp the outer edges and squeeze to hold it tightly. But for a soft object, like a sponge or the dolphin, the robots "learned" that it was easier to place one gripping "finger" in the middle and one around the edge, and then just squeeze.
The good news -- the Robots still needed humans to place objects in the trays.
Brown University project called “RoboBrain,” plans to store all machine-learned behaviors into a knowledge database which other bots can tap into. The project has been nicknamed “the world’s first knowledge engine for robots.” Some people call in "SkyNet."
Robo Brain is a large-scale computational system that learns from publicly available Internet resources, computer simulations, and real-life robot trials.
It accumulates everything robotics into a comprehensive and interconnected knowledge base. Applications include prototyping for robotics research, household robots, and self-driving cars. 

The Claw Is Our Master: He chooses who will go and who will stay.
Regarding robots walking into a bar... you would think that 14 robots using convolutional neural networks would be great at telling jokes. They're not.
The best robot-written humor I've found is derivative of human jokes. For example, the CMU bot, dubbed Data, began a recent robot standup session with,

“A doctor says to his patient, ‘I have bad news and worse news. The bad news is that you only have 24 hours to live.'” The patient wondered how the news could possibly get worse. “I’ve been robo-dialing you since yesterday...”
More on robots, AI and humor, here: It's hard work being funny--especially for robots.
Click this link for more information regarding Deep Learning and  Convoluted Neural Networks.
(c) David J. Katz, 2016 - New York City


Popular posts from this blog

Annotated Guide To Men's Belts

The Complete Guide To Men’s BeltsArticle By  on 11th March 2014 | @gabrielweil


Stay Hungry. Stay Foolish.

Nine years and 19 million YouTube views later, Steve Jobs's commencement address to Stanford University's graduating class of 2005 has achieved iconic status. Jobs, the Apple visionary who died in 2011 at age 56, delivered a speech that resonated far beyond the Stanford audience, with a masterful mix of personal anecdotes, sparks of insight and universally applicable pieces of wisdom. Each year, especially around graduation season, people discover and rediscover Jobs's speech and its messages for those who seek meaning and purpose in life and at work. - Carolyn Gregoire Note that Steve Jobs originally asked Screenwriter Aaron Sorkin to write this speech: Sorkin was not available.  - DJK Full text of Steve Jobs' commencement address to Stanford University 2005
"I am honored to be with you today at your commencement from one of the finest universities in the world. I never graduated from college. Truth be told, this is the closest I've ever gotten to a college gradu…

3D Printed Dinner & Neckwear

Dinner Is Printed By A. J. JACOBS - New York Times
THE hype over 3-D printing intensifies by the day. Will it save the world? Will it bring on the apocalypse, with millions manufacturing their own AK-47s? Or is it all an absurd hubbub about a machine that spits out chintzy plastic trinkets? I decided to investigate. My plan: I would immerse myself in the world of 3-D printing. I would live for a week using nothing but 3-D-printed objects — toothbrushes, furniture, bicycles, vitamin pills — in order to judge the technology’s potential and pitfalls.
I approached Hod Lipson, a Cornell engineering professor and one of the nation’s top 3-D printing experts, with my idea. He thought it sounded like a great project. It would cost me a mere $50,000 or so. Unless I was going to 3-D print counterfeit FabergĂ© eggs for the black market, I’d need a Plan B. Which is how I settled on the idea of creating a 3-D-printed meal. I’d make 3-D-printed plates, forks, place mats, napkin rings, candlesticks —…